Atlas Daedalus (acode)

Industrial Autonomous Terminal AI Coder · Sovereign Pure Rust Engine · Atlas Research Labs

Atlas Daedalus is an industrial-grade, terminal-native AI software engineering agent designed to supersede heavy, cloud-locked coding assistants (Claude Code, Cursor CLI, OpenAI Codex CLI).

Built in 100% pure native Rust with zero V8 runtime overhead, Daedalus executes with sub-millisecond cold start times, embedded Ripgrep search, automated test self-healing loops, and strict zero-emoji OLED monochrome precision.

Technical Specifications

| Metric | Claude Code / Cursor CLI | Atlas Daedalus (acode) | |---|---|---| | Engine Architecture | TypeScript / Python / Node.js | Pure Native Rust (x86_64, aarch64) | | Memory Footprint (RSS) | 150MB - 350MB | < 8MB RSS | | Cold-Start Latency | 350ms - 800ms | < 5ms | | Codebase Search | Spawns external shell / regex | Embedded Ripgrep Engine (ignore::WalkBuilder) | | File Diff Generation | Spawns external git process | Native In-Memory (similar) Unified Diffs | | Automated Self-Healing | Brittle / prompt-wrapped | Autonomous loop (cargo test, npm test, pytest) | | Platform Coverage | Desktop only | Windows, Linux, macOS, Android (Termux) | | Interface Philosophy | Cartoonish / bloated emojis | Strict Zero-Emoji OLED Monochrome ANSI |

Universal Installation

Daedalus ships with pre-compiled static native binaries for Windows, Linux, and Android (Termux aarch64). Rust is NOT required on client devices.

Method 1: NPM / NPX (Zero Setup)

Run instantly without global installation:

# Interactive REPL
npx acode-cli

# Autonomous task execution
npx acode-cli "implement jwt authentication and verify tests pass"

Or install globally across any platform:

npm install -g acode-cli
acode

Method 2: Android Termux (Mobile)

On Android Termux, install Node.js and run directly:

pkg install nodejs
npm install -g acode-cli
acode

The package automatically executes the pre-compiled linux-aarch64 native engine with zero compilation needed.

AI Providers & Authentication

Daedalus connects seamlessly to sovereign AI routers, multi-model gateways, and local backends.

Configure credentials in ~/.daedalus/.env or via standard environment variables:

# Atlas AI-Powerbox (Sovereign Gateway)
export POWERBOX_API_KEY="pbk_live_..."
export POWERBOX_URL="https://powerbox.atlasresearchlabs.online"

# Google Gemini (Default: gemini-2.5-flash)
export GEMINI_API_KEY="AIzaSy..."

# Anthropic Claude (Default: claude-3-5-sonnet)
export ANTHROPIC_API_KEY="sk-ant-..."

# OpenAI (Default: gpt-4o)
export OPENAI_API_KEY="sk-proj-..."

# Local Ollama (100% Offline Sovereign Inference)
export OLLAMA_HOST="http://localhost:11434"
acode -p ollama -m deepseek-coder:6.7b

Interactive REPL Slash Commands

Inside the interactive terminal session, invoke slash commands for instant developer actions:

| Command | Action | |---|---| | /help | Print cheatsheet and keyboard navigation keys | | /diff | Display color-coded unified git diff of all modifications | | /status | View modified files, token counters, and active engine configuration | | /commit | Generate conventional commit and commit with sovereign author | | /cost | Display session token consumption, network latency, and cost ledger | | /audit | Execute autonomous AST static code review | | /tunnel | Expose active port via Beacon public tunnel | | /pulse | Inspect host CPU, memory, and process health metrics | | /model <m> | Dynamically switch LLM model mid-session | | /clear | Clear terminal viewport and reset trajectory buffer | | /exit | Terminate session |

License

MIT (C) 2026 Atlas Research Labs